Enhancing Evolutionary Optimization in Uncertain Environments By Allocating Evaluations Via Multi-Armed Bandit Algorithms

2. The computer-implemented method of claim 1, wherein the multi-arm bandit algorithm is a variant of an upper confidence bound (UCB) algorithm that determines the performance score for each candidate funnel dependent upon an exploration factor and an exploitation factor.

3. The computer-implemented method of claim 2, wherein the variant UCB algorithm performs a scoring cycle T times, T being a predetermined integer greater than zero.

4. The computer-implemented method of claim 1, wherein the multi-arm bandit algorithm is a Thompson Sampling for Bernoulli Bandits algorithm that determines the performance score for each candidate funnel dependent upon a probability density function of a beta distribution taking into account successes and failures of each candidate funnel.

5. The computer-implemented method of claim 4, wherein one variable of the beta distribution relates to successes of candidate funnels and another variable of the beta distribution relates to failures of candidate funnels.

6. The computer-implemented method of claim 1, wherein the multi-arm bandit algorithm is a Successive Rejects algorithm that reduces the candidate funnels to a single candidate funnel.

7. The computer-implemented method of claim 1, wherein the fitness score (F) is determined by F=sx/nx.